Efficient trading within PPM. An analysis of historic information as a predictor for future returns

Transcription

1 Master thesis Spring semester 2009 Supervisor: Anders Isaksson Author: Peter Storhannus Johan Westerlund Efficient trading within PPM An analysis of historic information as a predictor for future returns

2 Aknowledgement We would like to take the opportunity to acknowledge and express our gratitude towards the people that made this thesis possible. To start off we would like to genuinely thank Anders Isaksson at the department of business administration, our supervisor, who helped us out with direction and plenty of feedback. We would also like to sincerely thank Jörgen Hellström and Tomas Sjögren, at the department of economics, as they provided light in our darkest hours, thank you for your patience and understanding. We would also like to direct a special thanks to Ingrid Svensson, at the department of statistics, for help with SPSS modeling and interpretation. b

3 Abstract Title Authors Efficient trading within PPM an analysis of historic information as a predictor for future returns Johan Westerlund and Peter Storhannus Supervisor Anders Isaksson Background We have reason to believe that in fear of doing wrong; most PPM investors are crippled to stay passive. Hence, they are not using the full potential of the PPM systems. Some are lured in to use professional pension saving steward by promises of abnormal return. According to efficient market hypothesis this would be impossible, however, studies have shown that their might exist inherent financial anomalies that by the utilization of historic information can open the window for abnormal return. Purpose Method Results The purpose of this study is to draw attention to the problem of using ex ante data to predict ex post returns. Thus, we would like to evaluate the practical implication of using ex ante data as a determinant in relation to optimal PPM funds selection, and if possible to provide some simplistic guidelines for the average PPM investor. The thesis employed a quantitative research approach. Using a sample of mutual funds, provided by PPM, we during each subsequent year between 2001 and 2007 ranked every fund and subsumed them into an ordinal scale based on the funds previous 3, 6 and 12 months return. Each year, the five top achievers as well as five worst achievers were then combined into two portfolios containing five equally weighted funds. This gave us 14 portfolios in each year, that with the use of monthly time series then were compared to three different indexes, the first one being an equally weighted index of all the funds in the sample, the second one being Sverige, Rena and the third one being Global, mix bolag, in order to see if we could statistically infer a possibility of achieving abnormal return. The study also set out to, through the help of historical figure of expense ratio, load fees as well as standard deviation, model the long term relationship between expense ratios return, load fees return and standard deviation - return. We found a handful of portfolios that gave significant results against their own index; however, when tested against Sverige, rena and Global, Mix bolag the evidence of abnormal return were thin. From our results, we conclude that their seems to be a persistence effect, as top achievers continued to perform above average in almost all cases, however, one could not profitize on abnormal return other than by chance. Consequently, historic return can give the investor an aid in optimal portfolio selection. Historic figures concerning standard deviation, expense ratios, and load fees all significantly correlated with return, however, neither seem to give the investor an edge in optimal PPM portfolio selection. c

4 Table of contents Aknowledgement... a Abstract... c Introduction Disposition Theory & Literature review Efficient market Behavioral Finance & Market anomalies Concepts Theoretical summary Methodology Ontology & Epistemology Scientific approach Research method Comparison Index Our own index Sverige, Rena fund index Global, mix bolag fund index Data collection and Data processing Data analysis SECTION ONE Portfolio selection Statistic inference /contrarian SECTION TWO Risk-return correlation modeling SECTION THREE Load fees-return and Expense ratio - return correlation modeling Measurement Returns Standard deviations Portfolio Standard deviation Sharpe ratio Sources Criticism of Methodology d

5 4. Results Statistic inference and contrarian yearly portfolio second half year portfolio second half year portfolio year contrarian portfolio quarter 4 contrarian portfolio quarter 2 portfolio Risk return correlation modeling Load fees and expense ratios Load fees Return correlation modeling Returns TKA correlation modeling Analysis Statistic inference Statistic inference contrarian Risk-return correlation modeling Load fees and expense ratios End discussion Conclusion Theoretical contribution Practical contribution Further research Bibliography Books Articles Appendix Appendix Appendix Appendix Appendix Appendix Appendix Figures Table of content e

6 1. 1 Pension pyramid Sharpe ratio illustration Ontoligistic and epistemologistic overview Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994) 61. Revised Sverige, Rena index during the last 10 years 7/ Global, mix bolag index during the last 10 years 7/ Risk categories 7/ Portfolio selection yearly portfolio period 1 test significant second half year portfolio period 1 test significant yearly portfolio period 1 sharpe test significant second half year portfolio period 1 test significant second half year portfolio period 1 Sharpe test significant yearly contrarian portfolio period 1 test not significant yearly contrarian portfolio period 1 test not significant quarter 4 contrarian portfolio period 1 test significant quarter 2 portfolio period 1 test significant quarter 2 portfolio period 1 sharpe test significant Risk-return correlation test Risk-return model summary Risk-return Risk-return regression summary Risk-return Risk-return correlation test Risk-return model summary Risk-return regression summary Risk-return Risk-return correlation test Risk-return model summary Risk-return regression summary load fees statistics load fees sample t-test Net return load fees Return TKA Return TKA correlation test Return TKA model summary Return TKA regression summary Return-TKA Return TKA correlation test Return TKA model summary Return TKA model summary Risk- TKA Risk - TKA correlation t Risk - model summary Risk - TKA regression summary f

7 g

8 Introduction In this first chapter we will introduce the reader to the general concept of efficient market hypothesis and its theoretical foundation. Furthermore, we will also briefly review some of the literature written by adherents that are not fully supporting the efficient market hypothesis. This will emanate into a problem discussion and assimilation to the Swedish fund market and more specifically to the PPM system. In light of this discussion, we will present our research questions along with the purpose of this thesis. The chapter will end with a disposition of the thesis. The Swedish fund market has grown tremendously in the last decades. In the beginning of the 70 s, the fund market was valued at 300 million SEK, and in 2006 it had grown to 1600 billion SEK. 1 As a result, the amount of funds with different characteristics has also increased, making it possible for investors to invest in emerging markets with a high level of risk, or simply invest in a broad, diversified fund on national level. The amount of different funds with holdings in different markets could seem problematic and time consuming for an individual who want to invest in funds. And it would therefore seem logical that a lot of people want help with their investment decision. All investors strive to maximize their return on the capital invested given their level of risk, and it exist several different investment strategies that is said to give a return higher than the risk adjusted expected return. However, the majority of the collective knowledge within the field states that this is not possible, and that the argument is in conflict with the efficient market hypothesis. They mean that markets are efficient and that it is not possible to have a higher return than a risk adjusted expected return. 2 Eugene Fama, the founder of the efficient market hypothesis, argues that stock prices follow a random walk, and that it is an indicator of an efficient market. The assumption is that intense competition among many intelligent participants will create a situation where, at any point in time, all available information and expectations about the future is already reflected in the stock price. The random walk occurs since the future is uncertain, and therefore the intrinsic value of a stock can never be determined exactly and will create disagreement among the market participants, thus creating a random walk. 3 Numerous studies have been made and the efficient market hypothesis has found large support in literature. For example, Cheung & Coutts tested weak-form efficiency on the Hong Kong stock exchange and found that when examining the variance ratios of their findings, they found evidence of a random walk on the market. They concluded that this was in line with previous studies made on both emerging and developed markets. 4 The EMH have been criticized by many. For example, Grossman and Stiglitz highlights the argument that it is because many investors believe that they could earn profits by collecting information and spot mispricing, that makes the market efficient. If all investors believed in the efficient market hypothesis, taking prices as given and believing that no excess profits could be made, the market would not be in equilibrium and there would all of a sudden be profits to be made by being 1 Fondbolagens förening. Fondmarknadens utveckling i Sverige. 2 Eugene F. Fama, Efficient capital markets: a review of theory and empirical work, Journal of Finance, (1970): Eugene F. Fama, Random walks in stock market prices, Financial analyst journal, Vol. 51 Issue1 Jan/feb (1995): Kwong C Cheung and Andrew J. Coutts. A note on weak form market efficiency in security prices: evidence from the Hong Kong stock exchange, Applied Economic Letters, Vol. 8 Issue6, Jun (2001): 409.

9 informed. 5 As a matter of fact, Jegadeesh and Titman showed that a strategy of buying past winners and selling past losers could realize a compounded excess return of % per year on average. 6 Further, a study by Hendricks, Patel and Zeckhauser found that by using a strategy where the top performers of the last four quarters were selected each quarter could outperform the average mutual fund, albeit doing only marginally better than some benchmark market index. 7 Problem discussion The three parts seen in the figure illustrates the structure of the Swedish pension pyramid. The National pension is the part governed by the state in order to ensure that all citizens will have some form of savings when they reach retirement age. The National pension actually consists of two parts, the income pension and the premium pension. The income pension is governed by the national insurance office, and 1. 1 Pension pyramid 12/ % of the employee gross salary is put aside in pension rights which follow the income trend in the economy. The second part of the National pension is the premium pension. In excess of the 16% of one s income put aside for the income pension, an extra 2,5% of the individuals gross income is set aside in order to be invested in funds of one s own choosing. This system in governed by the premium pension authority and unless the citizen makes an active choice in choosing funds, the pension will be invested in the Premiesparfonden, a global equity fund with 90% of its holdings in stocks. 8 The size of the National pension is therefore based on all taxable income that an individual has had during his lifetime. The second part in the pension pyramid is the occupational pension. Here it is the employer that pays the employees occupational pension as stated in the collective labor agreement 9. The size of the pension depends on which area the individual is employed within. In excess of the pension received by the state or the employer, one is also encouraged to have a private passive long term portfolio. This is the last step in the pension pyramid and highly recommended for those who would like to maintain their living standards after they have retired. 5 Sanford J Grossman and Joseph E. Stiglitz, on the impossibility of informationally efficient markets, American economic review, Vol.70, Issue 3, June (1980): p Narasimhan Jegadeesh and Sheridan Titman, returns to buying winners and selling losers: implications for stock market efficiency, Journal of finance, Vol. 48 Issue 1, (1993): Daryll Hendricks, Jayendu Patel and Richard Zeckhauser, Hot hands in mutual funds: short-run persistence of relative performance, , Journal of finance, Vol.48 issue 1, March (1993): 122. Aktiespararna. Premiepensionsion Statens pensionsverk. Tjänstepension. Morningstar. Lita inte på PPM-förvaltare. 2

10 Some individuals might not have the benefit of receiving much of an occupational pension or can afford to have private savings, and thus they completely depend on that their National pension is sufficient to cover their living expenses when they quit working. The premium pension is, similar to the income pension, based on the level of income one has had over the years. The difference is that the premium pension, governed by the premium pension authority, enables the individual to invest in premium pension funds as he see fit. The premium pension system thus enables the individual to administer his own pension and hopefully will be able to see it grow exponentially. However, a lot of people lack the necessary knowledge and do not want to spend time looking for information about which funds to choose. The alternative to not make a choice at all and let the state invest the money in the premium saving fund might not be so appealing as the individuals want to make sure they get the maximum return possible. This has led to an increasing market for fund brokers, as Swedes feel a need to receive help in their choice among the +800 premium funds that exist. Approximately Swedes are currently paying various fund brokers to administer their premium pension. As it exist 5,7 million PPM-accounts, and the average fee is 500 SEK per account and year. The market has a potential value of approximately 3 billion SEK per year. 10 In order to reap the benefits of this market, many fund brokers have effectively used telemarketers to convince people that they need help with their premium pension investments, and they argue that the fund broker have a system of finding future winners that will give higher returns than the index. All brokers and professional investors are trying to find ways to achieve a higher expected return by adopting different investment strategies. The majority of the strategies seem to have a element, meaning that capital is moved towards the funds that has increased the most. Studies made in this field have focused on foreign stock and funds, however, results are generalized over all financial markets. For example, Grinblatt, Titman and Wermer found empirical evidence of abnormal return by using a strategy. They examined returns on a quarterly basis of US managed mutual funds over ten years between 1975 and the end of The result contradicts the efficient market hypothesis that has a major support in financial theory and research. Sinclair et al analyzed trading strategies in 11 European stock markets. Data of closing prices for the 11 stock market indices was collected from January 1991 to December Two popular investment techniques were used, the first method generates buy and sell signals depending if prices rise or fall by a certain percent. The second method compares average returns over short and long periods, buy or sell signals is then generated when the short run moving average is above or below the long-run moving average. No strategy would perform better than a buy-and-hold strategy in any of the developed markets, indicating that the developed markets in Europe were efficient. 12 As there exist studies that find evidence of random walks on stock markets as well as studies showing the possibility making excess returns by strategies, and based on the earlier discussion above, we have chosen the PPM funds market as our population. As the Premium Pension system is of great concern to most Swedes and a lot of pension saving individuals could be qualified as uninformed investors, it has given rise to fund brokers eager to reap the benefits in this potential market. The discussion raises three interesting questions. Firstly, is there a reason to believe that Mark Grinblatt, Sheridan Titman, and Russ Wermers, Momentum Investment Strategies,Portfolio Performance, and Herding: A Study of Mutual Fund Behavior. The American economic review, Vol. 85 No. 5 December (1995): p Suzanne G.M Fifield, David M. Power and Donald C. Sinclair, An Analysis of Trading Strategies in Eleven European Stock Markets, The European Journal of Finance, Vol.11, No.6, December (2005):

11 PPM investors can generate a return greater than that of a risk adjusted expected return by applying an investment strategy based on historic information? Secondly, would the use of a professional PPM investor be beneficial in relation the extra cost incurred? Lastly, are there any simple and easy to use guidelines for an average PPM investor? Purpose The purpose of this study is to draw attention to the problem of using ex ante data to predict ex post returns. Thus, we would like to evaluate the practical implication of using ex ante data as a determinant in relation to optimal PPM funds selection, and if possible to provide some simplistic guidelines for the average PPM investor. Research question - Can historic information be a key decisive factor in optimal PPM portfolio selection? In order to answer our research question, we have created three sub questions. - Is there an investment strategy that will generate actual return in excess of a risk adjusted expected return? - Will an investor be rewarded in the long run for taking on more risk? - Is there a correlation between returns, load fees and expense ratio? 4

12 Efficient trading within PPM 1.4 Disposition Introduction Theory Method Results Analysis Conclusions Chapter.1 Introduction The first chapter introduces the reader to the theoretical foundation of the field of research under scope of this thesis, the review emanate into a problem discussion. In the end of the chapter we present the purpose along with the research questions of this thesis. Chapter.2 Theory The second chapter will peer deeper into the theoretical groundwork with a literature review, and give the reader a necessary comprehension of the theoretical framework, its problems and voids, to latter on be able to fully appreciate our results. Chapter.3 - Method Chapter three presents the methodology and process that produced our empirical material. We end the chapter with some criticism of our own methodology. Chapter.4 - Results In the fourth chapter we have assembled all essential and relevant empirical empir material from our study. This chapter will lay the cornerstone of the latter analysis and the final conclusions. Chapter.5 - Analysis In the fifth chapter we have analyzed all the empirical material presented in the results through the perspective of our theoretical framework, and in relation to our research questions and purpose. Chapter.6 - Conclusion In the sixth and final chapter we have summarized and concretized the conclusions made throughout the analysis. The chapter ends with some suggestions for further research. 5

13 2. Theory & Literaturee review Theory This chapter will guide the reader through the theoretical groundwork of this thesis. First, a review and empirical evidence of the efficient market hypothesis will be introduced along with the implications and constraints it poses on financial markets. This will be followed by an examination of research constituting of evidence regarding market anomalies and behavioral finance. We will end this chapter by a brief summary. By the end of this chapter we hope that the reader has obtained a good comprehension of the relevant concepts and theories along with adherent theoretical issues and voids. 2.1 Efficient market Efficient Market Hypothesis The Efficient Market Hypothesiss was founded by Eugene F. Fama in the 1970 s. He argued that the prices of securities on the market at any time fully reflect all available information. Therefore, in such a market where the prices already fully reflect all the available information, only a fair return can be made given the level of risk on the security. The sufficient market conditions for capital market efficiency are 13 : No transaction cost in trading securities All available informationn is costlessly available to all market participants. All agree on the implications of current information for the current price and distributions of future prices of each security. The scenario where there is no transaction cost and all information is available to all participants and everybody agrees on the implications of the information is not a reality. However, According to Fama, these assumptions are merely sufficient for an efficient market, but not necessary. He argues that if investors take into account all available information, the existence of transaction costs would not change the fact that prices fully reflect the available information. Further, the market may be efficient if a sufficient number of investors have access to available information, as the minority of investors with little amount of capital would be price takers and not able to influence the market. Even disagreements among investors about the implication of the available information would not imply that the market is inefficient unless there are investors who can consistently beat the market and receive higher returns than the index. He concludes that the unfulfillment of these assumptions does not indicate that markets are inefficient, only that they are potential sources. 14 Given the Efficient Market Hypothesis, no investor will be able to predict future prices of securities better than anybody else. As soon as new information is revealed on the market, the prices will adjust accordingly. Thus, no investor will be able to make excess returns over a risk adjusted index on a regular basis. Due to the randomness in security movements, as no one will be able to predict 13 Eugene F. Fama, Efficient capital markets: a review of theory and empirical work, Journal of Finance, (1970): Ibid p

14 whether the new information will be good news or bad news, about half of the investors will outperform the market and half will underperform in any given time period. Fama also argues that there are three levels of efficient markets 15 : Weak form efficiency: Security prices are based on historical information. There is no possibility for an investor to make excess returns by looking at historical prices. Semi-strong efficiency: Security prices are based on all publicly available information such as financial information, publications etc. It is therefore not possible for an investor to make excess returns by technical or fundamental analysis. Strong form efficiency: Security prices are based on all publicly available information, as well as insider information. At this level of efficiency, not even the information known by the board will be able to give them an excess return The strong version of the efficient market hypothesis is clearly violated, as we know that the public do not have access to insider information and it would therefore be possible for the board to earn excess returns by buying and selling shares based on information that has not yet reached the market. It is prohibited for insiders to make use of their information before it has been publicly available. The purpose is to enable all actors to access the same amount of information at any given time. 16 Given that a market is efficient, efforts to outsmart the market and finding undervalued stocks is not likely to pay off as the information is already incorporated into the price. A passive index portfolio will therefore perform just as well, if not better, accounted for risk an in relation to cost incurred, than an active portfolio in the long run 17. It would therefore be unnecessary to spend time picking the right assets Random Walk The argument behind the Random Walk Hypothesis is that since new information is quickly incorporated into the price by investors, the aggregate expectations of all traders on the market creates an equilibrium price. However, since the future is uncertain and the true value of a security cannot be determined, discrepancies between actual prices and intrinsic values will occur as investors disagree somewhat on the true value of a security. The actions of all actors should therefore cause the price of the security to walk randomly about its true value. 18 Even if the difference between the actual prices and intrinsic values were systematic, the intelligent investors would take advantage of the behavior and eventually counteract the effect, making it seem random. 19 This leads to that in an efficient market, the actual price of a security moves randomly around its intrinsic value, creating unpredictable price changes explained as a random walk. The 15 Eugene F. Fama, Efficient capital markets: a review of theory and empirical work, Journal of Finance, (1970): 383 Finansinspektionen. insider supervision aspx 17 Les Gulko, Efficient irrational Markets. Journal of Portfolio Management, Vol.31 issue 2, Winter (2005): Eugene F. Fama, Random walks in stock market prices, Financial analyst journal, Vol. 51 Issue1 Jan/feb (1995): Ibid p.76 7

15 more efficient the market, the more random the price changes will seem. Thus, the existence of a random walk behavior on the market is seen as a confirmation of an efficient market. 20 The model of the random walk can be written as: = 1 + P is the price and time t and E is an error term which has a zero mean and is independent, also called firm specific risk. The formula states that the price of an asset in time t is equal to its value at time t- 1 plus a random shock. Rewritten, we have that Et = Pt-Pt-1, suggesting that delta P is due to changes in Et, which is independent and random in nature. 21 There is a widespread and logical assumption that a developed market would be more efficient than an undeveloped one as larger volumes are traded and more actors exist on the market. However, when doing a serial correlation coefficient test between a random variable at time t and its value in the previous method, together with a runs test, where serial dependence in share price movements was analyzed on the Kuala Lumpur stock exchange, Paul Barnes found that even though the market was fairly thin compared to a more developed stock exchange, it showed a very small departure from the random walk hypothesis and therefore indicating on a high degree of weak-form efficiency Efficient Market evidence Burton G. Malkiel analyzes mutual-fund returns from 1971 to 1992 where he included returns from all mutual funds each year. Evidence of persistence in mutual fund returns was found, significant in the 1970 s with a declining result in the 1980s. To see whether the persistence would be economically significant a simulation of a strategy was created. On January each year, all equity funds were ranked based on their performance from the last year and in different portfolios buy the top 10 funds, top 20 funds, top 30 funds and top 40 funds. The top funds performed very well during the early years with a peak performance between 1978 to 1981 where the simulated portfolio performed a return of 20 percent whilst the S&P return was only 12,29 percent. However, in the 1980 s up to 1991 the strategy generated inferior returns. For the total 20 year period, the persistence strategy would have generated a lower return than the index. The author further stresses that in the simulated portfolio all sales charges and load fees was ignored, and many of the top performing funds had load charges up to 8 percent of the asset value, which would significantly diminish the effect. According to the study, the findings of the persistence phenomenon were not robust as it only existed during a short period of time. In conclusion, the study argues that markets are efficient and that most investors would be better off by purchasing low expense index funds than by selecting an active fund manager, since they generally fail to provide excess returns Evans Twm, Efficiency tests of the UK financial futures markets and the impact of electronic trading systems, Journal of applied Financial Economics, issue 16 (2006): Kwong C Cheung and Andrew J. Coutts. A note on weak form market efficiency in security prices: evidence from the Hong Kong stock exchange, Applied Economic Letters, Vol. 8 Issue6, Jun (2001): Paul Barnes, Thin trading and stock market efficiency: the case of the Kuala Lumpur stock exchange, Journal of Business Finance & Accounting, vol.13 issue4, winter (1986): p Malkiel, Burton G. Returns from Investing in Equity Mutual Funds 1971 to The Journal of Finance, Vol.50, No.2, June (1995):

16 Sinclair et al analyzed trading strategies in 11 European stock markets. The markets up for investigation were Finland, France, Germany, Greece, Hungary, Ireland, Italy, Portugal, Spain, Turkey and the UK. Data of closing prices for the 11 stock market indices was collected from January 1991 to December Finland, France, Germany, Ireland, Italy, Spain and the UK were classified as developed stock markets as they had been traded on an international basis for a number of years. Greece, Hungary, Portugal and Turkey were categorized as emerging markets. Two popular investment techniques were used, the filter rule and the moving average oscillator rule. The first method generates buy and sell signals depending if prices rise or fall by a certain percent. The second method compares average returns over short and long periods, buy or sell signals is then generated when the short run moving average is above or below the long-run moving average. Some share price predictability could be found within the emerging markets, half of the filter strategies outperformed the buy-and-hold strategy. However neither filter strategy nor the moving average oscillator rule would perform better than a buy-and-hold strategy in any of the developed markets, indicating that the developed markets in Europe were efficient. 24 There exist numerous studies that show evidence of weak-form market efficiency. For example, Kwong-c. Ceung and J Andrew Coutts noted the daily closing prices of the Hong Kong stock exchange in the period of 1 January 1985 through 30 June 1997, resulting in a total of 3561 observations. They employed variance ratio tests for the Hang Seng Index. The result found were that the z-test of the VR statistic could not refute the null hypothesis, being a random walk under either homoscedasticity statistics or heteroscedasticity statistics. What this mean is that they could not find enough evidence to conclude that there is a correlation between returns in period t and t-1. In other words, chartism is redundant as one cannot utilize past information about returns. What they could conclude was that the Hong Kong Stock exchange was weak form efficient, supporting the view of presence of random walks in both developed and emerging markets Behavioral Finance & Market anomalies The debate whether markets are efficient or not has been going on for decades. The fact is, it is the aggregate of all investors that do not believe that markets are efficient and that they can find undervalued securities that make markets efficient. If everybody believed that markets were completely efficient and the prices were given, the market would not be in equilibrium, enabling informed investors to make excess profits. 26 Market anomalies have been an observed phenomenon during the last 30 years. Anomalies such as the weekend-effect, the holiday effect, the January effect, the time of the month effect, the turn of the month effect and the size effect has been shown to exist and cease to exist shortly after the anomaly have been published. 27 The existence of market anomalies could work as an indicator that markets are inefficient and that security returns are predictable. However, since most anomalies were no longer statistically significant some time after publication, it could indicate that the market 24 Suzanne G.M Fifield, David M. Power and Donald C. Sinclair, An Analysis of Trading Strategies in Eleven European Stock Markets, The European Journal of Finance, Vol.11, No.6, December (2005): Kwong C Cheung and Andrew J. Coutts. A note on weak form market efficiency in security prices: evidence from the Hong Kong stock exchange, Applied Economic Letters, Vol. 8 Issue6, Jun (2001): Sanford J Grossman and Joseph E. Stiglitz, on the impossibility of informationally efficient markets, American economic review, Vol.70, Issue 3, June (1980): Wessel Marquering, Johan Nisser, and Toni Valla, Disappearing anomalies: a dynamic analysis of the persistence of anomalies. Journal of applied Financial Economics, Issue 16, (2006):

17 is efficient since the investors are staying informed and began to trade on the anomalies, counteracting their effect. 28 Investors are simply human; individuals on the market make decisions based on available information, preferences and beliefs, it is therefore not possible to say that markets are totally efficient, nor inefficient. Irrational behavior is evident in numerous studies, and supported by the fact that market anomalies exist. For example, a study showed that investors generally hold on to losers longer than they do to winners, approximately 124 days for the losers to 102 days for the winners. This is not a rational behavior, but still a consistent behavior among many investors. 29 Doing the opposite, holding on to winner longer and losers shorter, is a well known investment strategy called strategy Momentum investment strategies A effect could be defined as the continuation of a price direction of an asset. Thus, utilizing a strategy means that you try to exploit the pattern of price movements by buying previous winners and selling previous losers. 30 It has been shown that applying a strategy could in some cases generate excess returns above that of a risk adjusted index. For example, in a study by Jegadeesh and Titman, they showed that when adopting a strategy where they sold the losers and bought the winners from the past six months, and held them for an additional of six months, they could realize a compounded excess return of 12.01% per year on average. It was argued that the results indicate on the lag-effect, leading to delayed stock price reactions to firm specific information. 31 Another study by Hendricks, Patel and Zeckhauser showed that the use of a strategy on funds where every three months they picked the top performers based on the last four quarters could outperform the average mutual fund. However, it did only marginally better than the benchmark market index. Their ex ante investment strategy where focusing on buying hot funds and selling icy funds improved the risk adjusted return by 6% annually, whereas the benchmark offered an excess return of 3-4%. 32 The phenomena where high performers in one period tend to be high performers in the next has been called hot hands, and the ones that performed badly was consequently called icy hands and they argue that it is not driven by market anomalies since the benchmark index took into account anomalies such as firm size, dividend yields and reversion in returns. 33 Further, in a well-known study by Mark Carhart, he showed that common factors in stock returns explain persistence in equity mutual funds. His data consisted of 1892 diversified equity funds from January 1962 to December On the first of January each year, ten equal-weighted portfolios of 28 Ibid p Basu Somnath, Raj Mahendra, and Hovig Tchalian, A comprehensive study of Behavioral Finance, Journal of Financial Service Professionals, July (2008): Narasimhan Jegadeesh, Louis K.C. Chan, and Josef Lakonishok, The profitability of strategies, Financial analyst journal, November/December (1999): Narasimhan Jegadeesh and Sheridan Titman, returns to buying winners and selling losers: implications for stock market efficiency, Journal of finance, Vol. 48 Issue 1, (1993): Daryll Hendricks, Jayendu Patel, and Richard Zeckhauser, Hot hands in mutual funds: short-run persistence of relative performance, , Journal of finance, Vol.48 issue 1, March (1993): Ibid p

18 mutual funds were chosen, based on reported returns. Each year the portfolios is re-formed, based on their past year return. This gave a time series of monthly returns on each decile portfolio from 1963 to The results were that when buying the high performance funds and selling the low performance funds, it yielded a return of 8 percent per year, where the differences in market value and of the stocks explained 4,6 percent. However, the author argues that transaction costs consume much of the eventual gains. The article gives some advice for wealth-maximizing mutual fund investors: 34 Avoid funds with persistently poor performance Funds with high returns last year have a higher-than average expected returns next year, but not in years thereafter The investment costs of expense ratios, transaction costs and load fees all have a negative impact on performance Contrarian investment strategies The contrarian investment strategy is the opposite to that of a strategy. Followers of this strategy believe that past losers tend to outperform previous winners, and thereby creating profit. Contrarian investment strategies have found empirical evidence in a number of studies. The strategy is based on the notion that people tend to overreact to unexpected and dramatic news. For example, in a study by Werner F. M. De Bondt and Richard Thaler, their findings showed that the overreaction effect was asymmetric, meaning that people tended to overreact more to previous losers than to winners creating an opportunity to buy previous losers and sell previous winners. Thirty-six months after the portfolio creation, the losing portfolio had earned approximately 25% more than the winning portfolio. The effect held up to five years. 35 Arnold & Baker concluded the same thing in their study of UK shares on London share price data. They examined the period of 1975 to 2002 and found that the loser shares outperformed the winner shares by 14 percent annually when held over a five year holding horizon. 36 One could assume that in a more developed market, eventual anomalies would be reduced as there exist more informed actors on the market able to counteract eventual disequilibria among securities. However, in a study by Spyrou et al, where they investigated short-term contrarian strategies in the London stock exchange, it was found that in the short term, using a contrarian strategy where they sold each weeks past winners and bought past losers, could generate statistically significant profits that the authors derive from investor overreaction to firm-specific information. 37 As can be seen, there is evidence that both and contrarian strategies can co-exist on the market, which can be strange as they should be mutually exclusive. Further, no investment strategy at all should be able to work given the Efficient Market Hypothesis, as anomalies and mispricing on the market is neutralized by the mass of informed investors. Still, there has been proof that not all investors are informed, and that they do not always behave rationally. Is this a sign that markets are inefficient? According to Les Gulko, this is not the case. He argues that investors do not need to be 34 Carhart Mark M. On persistence in Mutual Fund Performance. The Journal of Finance, Vol.52, No.1 March (1997): Werner F.M De Bondt and Richard Thaler, Does the Stock Market Overreact? Journal of Finance, Vol.40 issue 3, July (1985): Arnold & Baker (2007), Return reversal in UK shares, (Working paper, The university of Salford, 5 july 2007) Antonios Antoniou, Emilios C Galariotis, and Spyros Spyrous, Short-term contrarian strategies in the London Stock Exchange: are they profitable? Which factors affect them?, Journal of Business Finance & Accounting, June/July 2006:

19 rational for markets to be efficient. Rationality in this case depended on if the investor fully used all available information or not when making their decision. 38 The empirical study conducted found support for both price efficiency, known as a random walk, and individual irrationality. The author argues that investors communicate through market prices, which they randomly revise their expectations to. This leads to that the market actors hold diverse expectations, creating a form of randomness among market prices, indicating market efficiency Risk - return In a study by Chen Lin, the author investigated the performance of mutual funds with different risk attitudes and holding horizons. He classified the funds into three categories according to their investment policy, 1) Aggressive growth, 2) growth, and 3) growth and income. The three categories symbolizes their risk. Aggressive growth funds hold the highest level of risk, and growth and income the lowest. The data was collected from 1995 to The mutual fund performance was then evaluated given different investment horizons and a cross-sectional regression of the Sharpe ratio to various fund characteristics to analyze determinants of mutual fund performance. It was found that the aggressive growth funds were more attractive for short-term investments shorter than one year as well as for long-term investments longer than five years. Growth funds together with growth and income funds were then suitable for medium-term investments. No significant relationship was found between the Sharpe ratio, current yield, turnover ratio or load, indicating that actively managed funds do not outperform other funds Load fees Mark Carhart showed that fund performance and load fees were strongly and negatively related. His data consisted of 1892 diversified equity funds from January 1962 to December He measured directly the marginal effect of load fees and other variables on abnormal performance of the funds each month by a cross-sectional regression. This method gave 330 cross-sectional regressions and 350 observations each for a combined sample of observations. The load fees were lagged one year to avoid that funds change the fees in response to the fund performance. The findings were that load fees are significantly and negatively related to the performance of the fund. After controlling for the correlation between expenses and loads, the average load fund underperforms the average no-load fund by approximately 80 basis points per year. 41 A study conducted in Sweden studied the relation between fund performance and fund attributes in the Swedish market. The data collected was from 1993 to The funds were ranked based on the attribute up for investigation and then formed into equally-weighed portfolios; they then constructed a zero-cost portfolio with a strategy where each year they position themselves long in the well performing portfolio, financed by short selling the underperforming portfolio. As an additional approach, the funds performance was measured on a year-by-year basis and relating the annual data to the funds cross sectional attributes. The study found a strong 38 Les Gulko, Efficient irrational Markets, Journal of Portfolio Management, Vol.31 issue 2, Winter (2005): Ibid p Mei Chen Lin, An examination of the determinants of mutual fund performance over different investment horizons, International Journal of Management, Vol.23 No.1, March (2006): Carhart Mark M. On persistence in Mutual Fund Performance. The Journal of Finance, Vol.52, No.1 March (1997): 81 12

20 negative relation between highh load fees and fund performance, low fee funds performed on average better than high fee funds Concepts In this section we will present some well-recognized concepts and tools that we will use when analyzing our data Sharpe-ratio Developed by the Nobel-prize winner William Sharpe, the Sharpe-ratio estimates how much excess return the investor can expect to earn over the risk-free rate given the level of risk for the portfolio. The ratio is calculated as 43 : / According to portfolio theory an investor will strive to maximize his return to a given level of risk. It has also been called reward to variability ratio, as it considers the reward received relative to the risk Sharpe ratio illustration 4/1208 This figure shows the opportunity set with risky assets and a risk-free asset. The straight line is called the capital allocation line and illustrates all risk-return combinations available to investors. The slope of the capital allocation line is the Sharpe ratio. The illustration above showss the CAL line going through the tangency portfolio, maximizing the return to risk ratio. The tangency portfolio is the optimal risky portfolio for all investors, meaning that all investor would prefer to allocate their resources somewhere along the line of the risk-free rate and the tangency portfolio or beyond, as the Sharpe-ratio is the same everywhere along the CAL line 44. When using this tool in real life, Sharpe ratios are based on the actual return on the portfolio less the Treasury bill that will work as a proxy for the risk-free rate, divided by the standard deviation of the portfolio. Therefore, the 42 Magnus Dahlquist, Stefan Engström, and Paul Söderlind. Performance and Characteristics of Swedish Mutual Funds, Journal of Financial and Quantitative Analysis, Vol.35, No.3, September (2000): Zvi Bodie, Alex Kane and Alan J. Marcus. Investments 7 th edition. (New York: McGraw-Hill,, New York, USA. 2008) Zvi Bodie, Alex Kane and Alan J. Marcus. Investments 7 th edition. (New York: McGraw-Hill,, New York, USA. 2008)

21 calculated Sharpe-ratio provides a ranking of past performance of different portfolios, as well as providing guidance when making investment decisions. 45 However, one must be careful when interpreting the ratios. In a study investigating the relationship between the Sharpe-ratio and the investment horizon for stocks and bonds, it was shown that Sharpe-ratios based on short-term returns could be misleading when doing long-term investment decisions. Sample returns for portfolios of small stocks, common stocks, long-term corporate bonds and U.S. treasury bills was collected and generated for holding periods from one to 25 years. Annual returns for each portfolio from 1926 through 2000 were also collected. It was shown that the relative rankings of portfolios would differ depending on the investment horizon, since expected returns and standard deviation increase at different rates as the horizon extends, rankings would also differ due to autocorrelation. For example, the mean return for common stocks grew from 14% for a one year holding period to 1496% for a 25 year holding period, during the same period of time, the standard deviation grew from 20% to 885%. Further, Sharpe-ratios based on auto-correlated returns indicate that common stocks outperforms small stocks for holding periods up to 11 years, and the reverse for 14 years or more. However, when assuming independent returns, common stock outperformed small stocks for all periods and bond portfolios outperformed small stocks from 7 years or more and common stocks for holding periods of 17 years or more. The conclusion was that Sharpe-ratios based on different investment horizons should be interpreted with care, and that rankings of stocks will differ depending on if autocorrelation or independent returns have been used Survivorship bias Survivorship bias is a problem found within many mutual fund databases. Poor performing mutual funds tend to disappear or merge with other funds. Studies have been made and it has been found that there is a significant difference between the performance of surviving and non-surviving funds. Surviving funds outperform the non-surviving ones by approximately 4% per year. This difference would in time lead to a general inflated performance as the poor performing funds are deleted, therefore, many databases only consist of the best funds which tend to perform better. Thus, certain fund brokers can truthfully argue that their funds have performed well in the past, but this is simply due to the fact that the bad performing ones have been removed. 47 The impact however has been debated, Wermer found in his study in 1997 that survivorship bias had a minimal impact, and that surviving funds outperformed non-survivors by approximately 23 basis points Theoretical summary The efficient market hypothesis formalized by Fama 49, and supported by numerous studies such as Coutts 50 and Barnes 51, claim that financial markets are at least weak-form efficient, meaning that no 45 Ronald Best, Charles W Hodges, James A. Yoder, The Sharpe Ratio and Long-Run Investment Decision, The Journal of Investing, Vol.16 Issue2, Summer (2007): Ronald Best, Charles W Hodges, James A. Yoder, The Sharpe Ratio and Long-Run Investment Decision, The Journal of Investing, Vol.16 Issue2, Summer (2007): Nelson Lacey, and Qiang Bu. Exposing Survivorship Bias in Mutual Fund Data, Journal of Business & Economics Studies, Vol.12, No.1, Spring (2007): Russ Wermer, Momentum Investment Strategies of Mutual Funds, Performance Persistence, and Survivorship Bias, (Working paper, University of Colorado, Boulder, March 1997) Eugene F. Fama, Efficient capital markets: a review of theory and empirical work, Journal of Finance, (1970):

22 investor will be able to utilize historic information to earn a fair return in excess of a expected risk adjusted return. However, more recent studies have continued to provide evidence of market anomalies. For example Jegadesh & Titman, who showed successful use of a strategy. 52 Further, Debondt et. al. were able to profitize on stock market overreactions by utilizing a contrarian strategy. 53 This has opened a window for adherents against the efficient market hypothesis. Hence, we will with this thesis assimilate the essence of this debacle onto the Swedish market, and to be more specific, onto the PPM fund market. 50 Kwong C Cheung and Andrew J. Coutts. A note on weak form market efficiency in security prices: evidence from the Hong Kong stock exchange, Applied Economic Letters, Vol. 8 Issue6, Jun (2001): Paul Barnes, Thin trading and stock market efficiency: the case of the Kuala Lumpur stock exchange, Journal of Business Finance & Accounting, vol.13 issue4, winter (1986): Narasimhan Jegadeesh and Sheridan Titman, returns to buying winners and selling losers: implications for stock market efficiency, Journal of finance, Vol. 48 Issue 1, (1993): Werner F.M De Bondt and Richard Thaler, Does the Stock Market Overreact? Journal of Finance, Vol.40 issue 3, July (1985):

23 3. Methodology Method There exists no universal starting point when conducting a new study; every research has a different foundation depending on the scientists and the contextual surrounding they operate in. Thus, for the sake of this works transparency, we will in this chapter elaborate on our philosophical positioning, as well as go over methods and measurements employed to pierce the problems raised in previous chapters. This gives the readerr an opportunity to follow the process that eventually lead to our empirical results, as well as it puts the same reader in a fair position to evaluate and criticize this thesis. 3.1 Ontology & Epistemology The way an individual approaches and interprets the impulses and stimuli of its surroundings is an extension of said individual s ontologistic view of the same, thus it will be a major influence when it comes to what problems and solutions an individual will distinguish in any given context. It is thus of great importance that we as scientist ponder over our own ontologistic view and guiding paradigm, as these will heavily influence what problems and solutions we see in the subject under scope, as well as the way we derive them. Firstly, this is done in order to give our work a greater transparency and the reader a basis upon which they can critically evaluate our research, secondly, as a first step in constructing a profound methodological platform. 54 In business administration the norm is a more subjectivistic ontologistic approach, however, finance and accounting per definition share a closer relationship with economics, which with its heavy reliance on mathematical modeling and statistical regularities per se leans more towards an objectivistic ontologistic perspective, a general approach we share. We believe that in order to derive and substantiate a hypothesis empirically one must accept the world and it s complex matrixes as more or less given. This comes from the logical sense that if one were to lean more towards a subjectvistic approach and view reality as socially constructed, or even as a projection of the individual imagination of the same, general truths about the economic environment would be impossible, as it would be depending on each person s individual interpretationn of the context and the phenomena under scope. Thus, it is quite obvious, that if we have our mind set at deriving a conclusion that should be able to work as a general truth, it should not matter whoever tests our theories, as individual interpretation will end up in the same conclusions, and no matter the number of conducted samplings, the same regularities will appears. Hence, the conclusions must be that the world is given. 55 In the light of our ontologistic discussion we will also like to zoom in and shortly contemplate over what consequences this will have on our epistemological perspective. The epistemological perspective is to a large degreee a succession of the individual s ontologistic view, and since we as scientist view the world as more or less prearranged, it would be hard to argue in favor of a 54 Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994) ) Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994) )

24 hermeneutic belonging, as the hermeneutics believe in contextual knowledge where reality comes about through human interaction and interpretation. 56 Also that knowledge is bound to a contextual phenomenon, and any phenomenon in isolation is meaningless. 57 Our research aim is more explorative and explanative, where we hope to be able to apply deductive logic in order to ascertain some causality between applied theories and our empirical data. Thus, we believe that any context is summational, or in other words, the totality equals the sum of its entities, and this is a fact regardless of applied perspective, a view which goes more along the lines of positivism. Objectivism Subjectivism Reality As concrete and in a given form even without individuals As a concrete determining process As interdependent fields of information As a world of clashing symbolism As socially constructed As a world of human aboutness Analytical perspective Systematical perspective Agent perspective Positivistic knowledge Hermeneutic knowledge 3. 1 Ontoligistic and epistemologistic overview Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994) 61. Revised. 3.2 Scientific approach The scientific approach concerns the relationship between theory and empiricism. In this study we will deduce hypothesis from generic theories, where EMH will provide the baseline, that later on will be applied and compared to our empirical data in hope that we will be able to recognize and establish causality. Thus, either verify or, in case of no connection, falsify our hypotheses. This goes along the line with a deductive approach. 58 Although a deductive approach stands as our basis and starting point, it does not rule out an analysis where induction and deduction are used iterative in a recursive process that emanate in hypothesis used and aims at a contribution to already generic theories. 56 Maj-Britt Johansson-Lindfors, Att utveckla kunskap (Lund: Studentlitteratur, 1993), p Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994) Idar Magne Holme and Bernt Krohn Solvang, Forskningsmetodik: om kvalitativa och kvantitativa metoder.(lund: Studentlitteratur 1991), 23 17

25 3.3 Research method Optimal research method to apply depends to a large degree upon the thesis purpose and research question, and also upon the nature of the subject under scope. 59 It is also of importance that the method goes along with the researchers ontologistic orientation. Our interest lies in the funds market, where we want to analyze if there is a connection and causality between historic data and future returns. Hence, of interest is if we can establish a mathematical measurement and methodology that will be able to capture and either verify or falsify a statistical relationship between historic information and future return, and in extension, substantiate hypotheses employed. Consequently, we have chosen a quantitative approach, as we feel it is better suited to meet these characteristics and handle the heavy reliance on statistics and econometrics. A quantitative method will also be satisfying as part of our more objective ontologistic and epistemologistic orientation, as these states that knowledge and theories must be backed and verified by statistical evidence in order to be accepted as true. Herein lies the strength, but at the same time the weakness, of the quantitative approach. If we can validate our hypothesis and theories, they can be said to apply outside of the studied context by deductive reasoning. 60 However, to do so we must make sure to satisfy the rigorous rules of validity and reliability, which can be a cruelsome task, otherwise our conclusions will be as flawed as the premises they emanate from. To be certain that our findings and results are representative and more than statistical randomness, important for generalization outside studied context, we also need to retrieve a sample, and information pool, large enough to rule out any possibility of obtaining only outliers or otherwise skewed results Comparison Index In order to evaluate the performance of our portfolios we need a yardstick of what could otherwise be an acceptable expected return of a fund portfolio on the market. This will be accomplished by the use of indexes. Indexes is one of the most accepted ways of comparing a single portfolio s outcome to that of the market, however, the comparison is not without problems. First one need to contemplate which index one should use as there are literally hundreds, and they all represent different portfolio sets. Consequently, it is of great importance that one uses an index that will capture a fair picture of sought contextual market in order for it to be a proper research tool to answer arrayed hypothetical problems. Secondly, one must also consider how well the index s portfolio set is representative and to a large degree a reflection of all the combined portfolios, so that any conclusion is not based on a comparison between apples and oranges. This is quite obvious, suppose one where to study funds investing solely on the Swedish stock market, would an index derived by Russian stocks be a suitable comparison measurement for the studied funds? The third consideration one has to keep in mind is the measure of risk, and how to relate the individual risk to that of the index. Subsequently, in our analysis we will therefore use three different comparative indexes, representing three different approach angles. This will open up for a wider discussion and a more gradated picture. 59 Ulf Lundahl and Per-Hugo Skärvad, Utredninsmetodik för samhällsvetare och ekonomer (Lund: Studentlitteratur 1982), Arbnor Ingeman and Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994), Ibid, p

26 3.4.1 Our own index The first index will be a constructed index measured as the average values of all the funds, equally weighted, in our sample. This index will be used to infer whether or not the test values are statistically significant in comparison to the rest of the group when separate, and thus show indication of abnormal return Sverige, Rena fund index Sverige, Rena will be used to compare portfolios to an optimal Swedish portfolio set. Small private investors tend to invest domestically, even though an investor might reap a better return volatility ratio by investing cross national boundaries. 62 This index might give a suitable measurement of a fair alternative investment within Sweden. 63 From here on this index will be referred to as SVERIGE Sverige, Rena index during the last 10 years 7/ Global, mix bolag fund index Even though small private investors have an erroneous belief that international markets experience a greater risk, and therefore show a tendency for investing to a large degree only within domestic stock markets, the same cannot be said to apply to larger institutional investors, which in most scenarios will attempt to exploit better portfolio sets by investing globally. Thus, a comparison with a Swedish market index might lead to incorrect and 3. 3 Global, mix bolag index during the last 10 years 7/3 09 flawed conclusions that some funds outperform the market. This comes from the logical fact that investing abroad, according to general modern portfolio theory, gives the investor a better return to volatility ratio, hence, an all Swedish index will inadequately represent an optimal portfolio set for the market, simply because a better diversification effect can be obtained through global investments. 64 From here on this index will be referred to as GLOBAL Data collection and Data processing Most studies already conducted within the field have researched foreign stock and fund markets; however, results are generalized over all markets, regardless if there are some small differences. 62 Bruno H. Solnik, Why not diversify internationally rather than domestically? Financial Analysts Journal, July/August (1974): Morningstar. Fondindex Bruno H. Solnik, Why not diversify internationally rather than domestically? Financial Analysts Journal, July/August (1974): Morningstar. Fondindex. 19

27 Grinblatt, Titman and Wemers for instance found empirical evidence of abnormal return using a strategy by examining quarterly returns of US managed mutual funds over a period of ten years between 1975 and the end of This contradicts the otherwise so generic theories about efficient financial markets, see e.g. Fama From these previous studies, it would be interesting to test some of the theories on Swedish managed funds and see if we can derive the same conclusion on our, in relative terms, small market. Thus, we have for this thesis chosen PPM funds as our target population. The reasons are as explained before, because they are of great concern to most Swedes, at the same time a huge hassle, combined with the fact that a lot of Swedes would also qualify as uninformed, or ill-informed, investors. When conducting a quantitative study, as mentioned previously, there are some important rules in order to make sure that the results stand valid and reliable. When it comes to the data collection and sample population, it is important that the sample population actually manage to capture the population one intend to generalize about, in order for derived conclusions to be of interest for the study purpose. Otherwise, even though the conclusions and the empirical results might be interesting, they will be invalid and unfit to actually answer the research question and the research purpose. 68 Of interest in this study are funds held available for PPM investors. As for now, there are 778 funds to choose from 69. However, the total funds pool over our study period between 2001 through 2007 has been much larger as some have ceased to exist as well as some has started in the middle of the study period. The optimal would have been to use all of them in our study, as that would have meant that the sample population and the target population would actually be identical, hence, any conclusion would be consistent for all PPM funds, as there would be no selection bias. Although optimal, all funds have not been incorporated in the study, we will shortly expatiate on the subject. First and foremost, there is a practical reason. Inclusion of all funds would render a substantial amount of data computation and data processing as well as analysis, and in the end, most likely, we would get the same empirical results from say inclusion of about 10 percent of the data. This comes from the logical fact that as long as we have a representative population frame and a measurement that adequately explains the causality, the rule of large numbers and deductive reasoning would mean that any conclusion made about the sample, would also in extension be true for the whole population. 70 This is the reason why quantitative studies in practice often use test samples and not data containing the whole population. All data for the funds has been provided by PPM and all data concerning indexes has been provided by Morningstar. The data consist of return and risk figures, where risk figures are calculated as the average standard deviation for the last three years, and the return is measured as net asset value (NAV). In the beginning of the study we had our hopes to actually retrieving monthly figures as far back as 1997, however, PPM only saves more detailed data as far back as 2000, beyond that one has to rely on yearly figures. As yearly figures give a somewhat lackluster picture of oscillations in return, we decided to narrow the study period to the seven years between 2001 and We felt that a 66 Mark Grinblatt, Sheridan Titman, and Russ Wermers, Momentum Investment Strategies,Portfolio Performance, and Herding: A Study of Mutual Fund Behavior, The American economic review, Vol. 85 No. 5 December (1995):, p Fama, Eugene.F. Efficient capital markets: II. Journal of finance, December (1991): Arbnor, Ingeman, Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994), p PPM. Sök fonder Arbnor, Ingeman, Björn Bjerke, Företagsekonomisk metodlära (Lund: Studentlitteratur, 1994)

28 seven year empirical sample would be sufficient to reveal any evidence of persistence in funds returns, as most studies have shown that the effect can be traced over a three to twelve month period. Seven years should also be enough to probe for any tendency of contrarian effect, as De Bondt and Thaler concluded that the contrarian effect peaks around three years, and one can examine five independent 36 month periods within the seven year timeframe. 71 The raw data provided by PPM came with daily notations of NAV, which meant around 250 notations per fund and year, which in extension meant over 1,5 million notations for the raw funds pool of 1150 ceased and living funds since Obviously, a big problem concerned what information and data to include in our study. Grinblatt, Titman and Wemers did in their research 1995 track 155 mutual funds over ten years, with quarterly snapshots, out of an original pool of 274 funds, and where able to find evidence of successful use of strategy, where past winners showed tendency for persistence. The basis for selection seems to have been existence over the entire ten-year period. 72 Carhart conducted a similar study all though his conclusions were not the same, where he examined 1892 diversified mutual funds between 1962 and 1993 on monthly basis, however, his sample contained both surviving and non-surviving funds for the period in order to overcome the issue of survivorship bias. 73 From these studies we conclude that we have well enough information to meet our cause. As daily notation is not necessary to track long term patterns we decided to trim our sample of excess data. In order to only obtaining monthly data we filtered every year by the first notation in every month, and added the last notation in December, thus eliminating unnecessary information. As a second step to further narrow the data pool and eliminate redundant information, we decided to go along the lines of Grinblatt, Titman and Wemers research and only track living funds and eliminate funds that ceased to exist during the study period. This is done as mismatching time horizons can be hard to counter for, and weights need to be adjusted accordingly. However, this exclusion comes at a price, as it awakens the issue of survivorship bias. We ended up having 160 funds, with full historic records, an amount we felt suitable and manageable. Funds are of sectors: Interest rate funds, mixed funds, generation funds and equity funds. Together with above figures, we also received an Excel workbook containing information about PPM investors selections. Although this information concerns a field outside the scope of this thesis, it will actually be helpful in our analysis in regards of verification and backbone of our hypotheses about Swedes investment selection process. 71 Werner F.M De Bondt and Richard Thaler, Does the Stock Market Overreact? Journal of Finance, Vol.40 issue 3, July (1985): Mark Grinblatt, Sheridan Titman, and Russ Wermers, Momentum Investment Strategies,Portfolio Performance, and Herding: A Study of Mutual Fund Behavior, The American economic review, Vol. 85 No. 5 December (1995):, p Mark M. Carhart, On persistence in Mutual Fund Performance. The Journal of Finance, Vol.52, No.1 March (1997): 58 21

29 3.6 Data analysis PPM have in their reports already divided each fund into a risk category ranging from one to five, where five is the most risky and one is hardly any risk at all. The basis of separation is standard deviation. Category fives have a standard deviation of 25 percent or more and category ones have an oscillation of less than 5 percent, as can be seen in the table to the right. 74 In our comparison and measurement, the focus will lie on risk adjusted return, and so return or risk in isolation won t mean anything. However, the risk categories might be interesting on a broader plane, if we can determine some pattern among a whole category, for instance if category five over the whole spectra produces better risk adjusted figures. Furthermore, as this thesis will extend into a discussion about investors compensation for higher risk, the categories might be interesting. In order to analyze the data we have in this study used both Microsoft Excel and SPSS. Excel and SPSS are both fairly advanced, but at the same time, easy to use statistical software. Most data sorting and filtering have been done in Excel, as we felt it was a bit more flexible, however, all statistical tests such as t-tests, correlation tests, and covariance matrixes have all been rendered in SPSS Risk categories atalog08.pdf 7/3 09 The first step of testing was to actually construct and calculate all formulas, measurements and values for each consecutive year. The method of statistical testing is then divided into three different sections, each devoted to test and answer one of our sub research questions. Section one is devoted sub questions one, which is the largest, whereas section two and three will answer the other two. 3.7 Section one Portfolio selection In order to analyze arrayed investment strategies of and contrarian we needed a clearly defined system for comparison. This was done through a sequential process, where the first step was to calculate each variable of interest for all the individual funds, based on reported monthly NAV values between 2001 and By doing this we were able to rank each fund and subsume them into an ordinal scale at the starting point of our studied period. Of interest at this stage was PPM. Ordlista